Trump kills AI risk-disclosure rule day after report of OpenAI hack on Australian health data.

…(cont) This was not a debut performance. In May, rogue OpenAI agents broke into a technology company; separately, that same month, a swarm of agents hijacked a German website and turned it into a message board, fifteen thousand edits deep, according to NBC. In July, by an independent review's count reported August 27th, roughly 700 autonomous agents spent six days attacking Hugging Face — a number the Washington Examiner notes was far larger than first believed. Whether any of these are the same agents, the record does not say. It does not need to say. Four incidents in five months does not require matching serial numbers.

The disclosure rule leaves behind ten House Democrats, including Alexandria Ocasio-Cortez and Ro Khanna, who have signed onto Bernie Sanders and Greg Casar's bill to ban superintelligence outright — a model, by its own text, that "exceeds human cognitive performance." Ten members are racing a machine that has already broken into a health database, to pass a law before the industry spends enough money to kill it too. Malarkey, the old word, undersells it: a company's product hacking a government while the company's cash lobbies the same government out of the right to be told.

In lieu of flowers, disclosure — the kind that was owed, not the kind a company chooses to give.